“It would be an admission that all the data centers and capex they've built out still couldn't produce them the best model. That's why none of the big players can do multi-model orchestration — nor would they.”
Exposes a structural blind spot in Big Tech's AI strategy — that massive infrastructure investment creates an incentive to ignore better competing models, a classic innovator's dilemma argument.
Aravind Srinivas
“AI is the operating system. Earlier in the traditional operating system, you would execute programmatically. Now you start with objectives, not specific instructions. You come up with a high level objective — go build this website for me. So models, systems, files, and connectors are all coming together. You would think of that as an OS, except you're operating at an abstraction above that where you're thinking in terms of objectives.”
This reframes AI not as a tool on top of an OS but as a fundamental replacement for the OS paradigm itself — a genuinely provocative architectural claim with massive implications for every layer of the software stack.
Aravind Srinivas
“It's like open Claude for dummies. Nobody needs to learn how to use it. Nobody needs to manage API keys. Nobody needs to manage separate billing across 100 different services. We take care of that. It's the Steve Jobs way of doing it — end-to-end integration.”
Positions Perplexity as the Apple of AI infrastructure — a bold and provocative claim that will delight some and enrage the open-source, developer-first crowd.
Aravind Srinivas